Discussion

Key results:
Unlike previous epidemiological studies we did not find any evidence of a significant difference in CRS disease status between active smokers and non-smokers (p=0.5938). The lower number of active smokers observed in both CRS subgroups may in part be a consequence of the higher percentage of patients with concomitant asthma diagnosis as outlined in Table 1. Active smoking appears however to have a significant impact on quality of life in both CRSsNP and CRSwNP+ phenotypes although the underlying mechanism remains debated in the common literature. Multivariable analysis has shown that the higher SNOT-22 scores demonstrated in CRS smokers remains significant even after adjusting for age, sex and asthma diagnosis (Table 4). The Minimal Clinically Importance Difference (MCID) value for SNOT-22 is 8.9, this being the smallest change in treatment outcome that an individual patient would term meaningful. Although it does not necessarily follow that smoking negates the effect of treatment, the mean higher SNOT-22 score (>10) in smokers underlies the significant impact of smoking on overall symptom burden.
There was also no strong demonstrable evidence that active smoking increases the likelihood of need for revision sinus surgery although analysis of a larger cohort with standardised operative technique would help clarify this further.
Interpretation:
CRES is the largest epidemiological study of CRS in the UK to date and is the first study since the UK Sinonasal Audit to collect data on patient reported symptoms and smoking status in the context of a confirmed CRS diagnosis. The majority of previous population based studies have reported positive associations between CRS prevalence and tobacco use.1,2The conclusions drawn by some of these studies are limited by their own methodology, as unlike CRES they relied on self-reporting of CRS diagnosis and hence are open to overestimation of true disease prevalence. Analysing the UK CRES data, we have failed to demonstrate any such positive association. We are not the first study to find a lack of association with Pilan et al5 in Sao Paulo finding no significant difference in CRS prevalence according to smoking status (p = 0.43), total pack years (p = 0.26) or following exposure to second hand smoke (p = 0.18). Min et al4 also confirmed CRS diagnosis through physical examination but failed to find an association between active and or former smoking status and CRS prevalence. A more recent study by Lee et al21 reporting on data from the Korean Health population survey (KNHANES) found an increased CRS among active smokers however on multivariable analysis that there was no overall significant difference between CRS prevalence and the patients smoking status in those patients aged 40 years and below. They did however note a similar finding to that recorded in the European GA2LEN1 study that the number of years smoked is significantly associated with CRS prevalence (increasing by 1.5% for every year in total smoking period).
Some studies have suggested an increasing prevalence of CRS with total number of years smoked.1,21 The results from Caminha et al22 are however contradictory, finding on multivariable analysis that Chronic Obstructive Pulmonary Disease (COPD) incidence and hence a likely surrogate for greater smoking history was not associated with a higher prevalence of rhinosinusitis symptoms.
Lachanas et al23 previously demonstrated that within a general ‘non-CRS’ population, smokers have higher SNOT-22 scores compared to non-smokers. It is clear from the CRES data that similarly all active smokers (both active CRS and control patients) had average higher SNOT-22 scores, although this was only statistically significant for active smokers with confirmed CRS (Table 3). This adds some weight to the argument that tobacco smoke may have an adverse effect on nasal outcome measures independent to whether the patient has underlying CRS. This finding has potential implications for epidemiological studies that rely on CRS self-reporting or questionnaire-based assessments without concurrent endoscopic CRS confirmation. These studies are vulnerable to overestimating CRS complaints within the smoking population as smokers appear more likely to have QOL nasal complaints and may perceive this incorrectly as CRS.
Revision sinus surgery rates remain high in the CRS population, evidenced from the National sinonasal audit five year follow up which demonstrated increasing revision rates, reaching 19.1% at 5 years; greatest in those patient with nasal polyps (20.6%)24. Previous CRES analysis demonstrated that 45% of CRS patients reported some form of surgical procedure whilst multiple surgical procedures were reported in 4% of CRSsNP patients and 23% of CRSwNP+ patients.25Interestingly the CRES smoking cohort reported lower numbers of surgical interventions compared to non smokers (Table 5) and analysis failed to find a statistical difference between smoking status and multiple surgeries. These results suggest active smoking may not be a significant risk factor for requiring multiple surgeries, however given the nature of data collection and the low comparative number of smokers versus non smokers this may not be truly representative. There are however multiple variables that may contribute to the number of operations a patient undergoes including the level of surgeon experience and selection bias on whom to operate in which being an active smoker could play a negative factor.
Previous studies have assessed the consequence of tobacco use on symptom control and rates of revision surgery. Wu et al26analysed revision sinus surgery rates in patients with CRSwNP and found on multivariable analysis that smokers had a significantly shorter time period (median 2.82 vs. 4.31 years) before further revision surgery was deemed necessary. A recent literature review by Reh at al27 reported conflicting evidence with respect to surgical outcomes and smoking, whilst earlier studies tended to demonstrate a deleterious effect more recent prospective studies have failed to find an similar association. These conflicting literature findings may in part be accounted for by differences in surgical intervention (e.g. polypectomy alone versus full clearance FESS) and by evolving changes in technique and instrumentation over the years. Interestingly Rudmik et al28 in their prospective study reported that active smokers with recalcitrant disease can experience similar benefits and improvement in quality of life scores following endoscopic sinus surgery as their non-smoking peers. There remains however a lack of studies looking at large numbers of high-volume smokers which may help to clarify this association further.
The CRES analysis has demonstrated a higher symptom burden in active smokers, with a mean difference in SNOT 22 scores greater than the MCID. As an observational study we are limited in our conclusions; however our failure to demonstrate an association between active smoking and higher reports of revision surgery would align with recent prospective studies concluding that surgery can be effective in smokers and should be considered as a treatment option.

Limitations

The CRES study design has certain limitations, firstly the data was self-reported and may therefore predispose to recall bias. Secondly the study only included one specific question related to current tobacco smoking, allowing us to determine whether the patient was an active smoker and if they were a mild to heavy user. The selected question did not identify whether patients were ex-smokers and did not seek to quantify ‘pack year’ history nor did it enquire as to the presence of other tobacco users in the household. We are therefore unable to adequately comment on whether smoking is an independent risk factor for developing CRS or comment on the possible role of second-hand smoke exposure in CRS prevalence. The degree of tobacco use was not evenly distributed amongst the CRES cohort with only 6-7% of patients reported smoking heavily (>20 tobacco products a day). The data must also be interpreted considering associated reporting bias relating to the quantity people reported smoking, which could be an under-representation. A further limitation of the study design meant that data collection did not allow for calculation of total years smoked, we are therefore unable to accurately comment on whether prevalence of CRS in smokers appears dose dependent.

Generalisability

CRES is a cross sectional UK based study incorporating a variety of the CRS population from across the country presenting to secondary care. The CRES study does not necessarily capture the whole CRS spectrum as mild sufferers may be managed by primary care alone and may therefore be underrepresented. Further because of the multifactorial nature of CRS it is difficult to assess the impact of one single factor on CRS pathogenesis in isolation. In contrast to other studies, CRS was diagnosed by ENT specialists according to accepted diagnostic guidelines (EPOS 2012)16, other existing studies have relied on self-diagnosis and or used different criteria making direct comparisons with the existing literature more complicated.